Estimating sensor-space EEG connectivity: Identifying best performing methods for functional connectivity in...

Area of Science:

• Neuroscience
• Signal Processing

Background:

• Functional brain connectivity (FC) estimation using electroencephalography (EEG) is crucial for understanding brain function.
• Variability in EEG preprocessing methods complicates the comparison and integration of FC findings across studies.
• Standardization of EEG preprocessing is needed for reliable FC measurement and reporting.

Purpose of the Study:

• To assess the impact of different preprocessing steps on EEG-FC estimates.
• To validate a reliable methodological pipeline for EEG-FC assessment using simulated data.
• To identify optimal preprocessing choices for robust EEG-FC analysis.

Main Methods:

• Simulated EEG-FC data with known ground truth connections were used.
• FC estimates were compared across various FC measures and preprocessing variations.
• Key preprocessing steps included re-referencing and epoching.

Main Results:

• Segmenting data into 40+ epochs (6s+ length) yielded the most accurate FC estimation.
• Reference Electrode Standardization Technique (REST) and common average re-referencing performed best with imaginary coherence and weighted phase lag index (wPLI).
• Magnitude-squared coherence performed optimally with Current Source Density (CSD) reference-free techniques.

Conclusions:

• Preprocessing choices significantly influence EEG-FC measurements, including referencing, epoching, and FC metrics.
• This study provides evidence-based recommendations for robust EEG-FC assessment.
• The findings support the development of standardized EEG preprocessing pipelines for FC analysis.